@depup/fast-xml-parser 5.5.6-depup.0

package/src/xmlparser/OrderedObjParser.js

@@ -0,0 +1,905 @@
+'use strict';
+///@ts-check
+
+import { getAllMatches, isExist, DANGEROUS_PROPERTY_NAMES, criticalProperties } from '../util.js';
+import xmlNode from './xmlNode.js';
+import DocTypeReader from './DocTypeReader.js';
+import toNumber from "strnum";
+import getIgnoreAttributesFn from "../ignoreAttributes.js";
+import { Expression, Matcher } from 'path-expression-matcher';
+
+// const regx =
+//   '<((!\\[CDATA\\[([\\s\\S]*?)(]]>))|((NAME:)?(NAME))([^>]*)>|((\\/)(NAME)\\s*>))([^<]*)'
+//   .replace(/NAME/g, util.nameRegexp);
+
+//const tagsRegx = new RegExp("<(\\/?[\\w:\\-\._]+)([^>]*)>(\\s*"+cdataRegx+")*([^<]+)?","g");
+//const tagsRegx = new RegExp("<(\\/?)((\\w*:)?([\\w:\\-\._]+))([^>]*)>([^<]*)("+cdataRegx+"([^<]*))*([^<]+)?","g");
+
+// Helper functions for attribute and namespace handling
+
+/**
+ * Extract raw attributes (without prefix) from prefixed attribute map
+ * @param {object} prefixedAttrs - Attributes with prefix from buildAttributesMap
+ * @param {object} options - Parser options containing attributeNamePrefix
+ * @returns {object} Raw attributes for matcher
+ */
+function extractRawAttributes(prefixedAttrs, options) {
+  if (!prefixedAttrs) return {};
+
+  // Handle attributesGroupName option
+  const attrs = options.attributesGroupName
+    ? prefixedAttrs[options.attributesGroupName]
+    : prefixedAttrs;
+
+  if (!attrs) return {};
+
+  const rawAttrs = {};
+  for (const key in attrs) {
+    // Remove the attribute prefix to get raw name
+    if (key.startsWith(options.attributeNamePrefix)) {
+      const rawName = key.substring(options.attributeNamePrefix.length);
+      rawAttrs[rawName] = attrs[key];
+    } else {
+      // Attribute without prefix (shouldn't normally happen, but be safe)
+      rawAttrs[key] = attrs[key];
+    }
+  }
+  return rawAttrs;
+}
+
+/**
+ * Extract namespace from raw tag name
+ * @param {string} rawTagName - Tag name possibly with namespace (e.g., "soap:Envelope")
+ * @returns {string|undefined} Namespace or undefined
+ */
+function extractNamespace(rawTagName) {
+  if (!rawTagName || typeof rawTagName !== 'string') return undefined;
+
+  const colonIndex = rawTagName.indexOf(':');
+  if (colonIndex !== -1 && colonIndex > 0) {
+    const ns = rawTagName.substring(0, colonIndex);
+    // Don't treat xmlns as a namespace
+    if (ns !== 'xmlns') {
+      return ns;
+    }
+  }
+  return undefined;
+}
+
+export default class OrderedObjParser {
+  constructor(options) {
+    this.options = options;
+    this.currentNode = null;
+    this.tagsNodeStack = [];
+    this.docTypeEntities = {};
+    this.lastEntities = {
+      "apos": { regex: /&(apos|#39|#x27);/g, val: "'" },
+      "gt": { regex: /&(gt|#62|#x3E);/g, val: ">" },
+      "lt": { regex: /&(lt|#60|#x3C);/g, val: "<" },
+      "quot": { regex: /&(quot|#34|#x22);/g, val: "\"" },
+    };
+    this.ampEntity = { regex: /&(amp|#38|#x26);/g, val: "&" };
+    this.htmlEntities = {
+      "space": { regex: /&(nbsp|#160);/g, val: " " },
+      // "lt" : { regex: /&(lt|#60);/g, val: "<" },
+      // "gt" : { regex: /&(gt|#62);/g, val: ">" },
+      // "amp" : { regex: /&(amp|#38);/g, val: "&" },
+      // "quot" : { regex: /&(quot|#34);/g, val: "\"" },
+      // "apos" : { regex: /&(apos|#39);/g, val: "'" },
+      "cent": { regex: /&(cent|#162);/g, val: "¢" },
+      "pound": { regex: /&(pound|#163);/g, val: "£" },
+      "yen": { regex: /&(yen|#165);/g, val: "¥" },
+      "euro": { regex: /&(euro|#8364);/g, val: "€" },
+      "copyright": { regex: /&(copy|#169);/g, val: "©" },
+      "reg": { regex: /&(reg|#174);/g, val: "®" },
+      "inr": { regex: /&(inr|#8377);/g, val: "₹" },
+      "num_dec": { regex: /&#([0-9]{1,7});/g, val: (_, str) => fromCodePoint(str, 10, "&#") },
+      "num_hex": { regex: /&#x([0-9a-fA-F]{1,6});/g, val: (_, str) => fromCodePoint(str, 16, "&#x") },
+    };
+    this.addExternalEntities = addExternalEntities;
+    this.parseXml = parseXml;
+    this.parseTextData = parseTextData;
+    this.resolveNameSpace = resolveNameSpace;
+    this.buildAttributesMap = buildAttributesMap;
+    this.isItStopNode = isItStopNode;
+    this.replaceEntitiesValue = replaceEntitiesValue;
+    this.readStopNodeData = readStopNodeData;
+    this.saveTextToParentTag = saveTextToParentTag;
+    this.addChild = addChild;
+    this.ignoreAttributesFn = getIgnoreAttributesFn(this.options.ignoreAttributes)
+    this.entityExpansionCount = 0;
+    this.currentExpandedLength = 0;
+
+    // Initialize path matcher for path-expression-matcher
+    this.matcher = new Matcher();
+
+    // Flag to track if current node is a stop node (optimization)
+    this.isCurrentNodeStopNode = false;
+
+    // Pre-compile stopNodes expressions
+    if (this.options.stopNodes && this.options.stopNodes.length > 0) {
+      this.stopNodeExpressions = [];
+      for (let i = 0; i < this.options.stopNodes.length; i++) {
+        const stopNodeExp = this.options.stopNodes[i];
+        if (typeof stopNodeExp === 'string') {
+          // Convert string to Expression object
+          this.stopNodeExpressions.push(new Expression(stopNodeExp));
+        } else if (stopNodeExp instanceof Expression) {
+          // Already an Expression object
+          this.stopNodeExpressions.push(stopNodeExp);
+        }
+      }
+    }
+  }
+
+}
+
+function addExternalEntities(externalEntities) {
+  const entKeys = Object.keys(externalEntities);
+  for (let i = 0; i < entKeys.length; i++) {
+    const ent = entKeys[i];
+    const escaped = ent.replace(/[.\-+*:]/g, '\\.');
+    this.lastEntities[ent] = {
+      regex: new RegExp("&" + escaped + ";", "g"),
+      val: externalEntities[ent]
+    }
+  }
+}
+
+/**
+ * @param {string} val
+ * @param {string} tagName
+ * @param {string|Matcher} jPath - jPath string or Matcher instance based on options.jPath
+ * @param {boolean} dontTrim
+ * @param {boolean} hasAttributes
+ * @param {boolean} isLeafNode
+ * @param {boolean} escapeEntities
+ */
+function parseTextData(val, tagName, jPath, dontTrim, hasAttributes, isLeafNode, escapeEntities) {
+  if (val !== undefined) {
+    if (this.options.trimValues && !dontTrim) {
+      val = val.trim();
+    }
+    if (val.length > 0) {
+      if (!escapeEntities) val = this.replaceEntitiesValue(val, tagName, jPath);
+
+      // Pass jPath string or matcher based on options.jPath setting
+      const jPathOrMatcher = this.options.jPath ? jPath.toString() : jPath;
+      const newval = this.options.tagValueProcessor(tagName, val, jPathOrMatcher, hasAttributes, isLeafNode);
+      if (newval === null || newval === undefined) {
+        //don't parse
+        return val;
+      } else if (typeof newval !== typeof val || newval !== val) {
+        //overwrite
+        return newval;
+      } else if (this.options.trimValues) {
+        return parseValue(val, this.options.parseTagValue, this.options.numberParseOptions);
+      } else {
+        const trimmedVal = val.trim();
+        if (trimmedVal === val) {
+          return parseValue(val, this.options.parseTagValue, this.options.numberParseOptions);
+        } else {
+          return val;
+        }
+      }
+    }
+  }
+}
+
+function resolveNameSpace(tagname) {
+  if (this.options.removeNSPrefix) {
+    const tags = tagname.split(':');
+    const prefix = tagname.charAt(0) === '/' ? '/' : '';
+    if (tags[0] === 'xmlns') {
+      return '';
+    }
+    if (tags.length === 2) {
+      tagname = prefix + tags[1];
+    }
+  }
+  return tagname;
+}
+
+//TODO: change regex to capture NS
+//const attrsRegx = new RegExp("([\\w\\-\\.\\:]+)\\s*=\\s*(['\"])((.|\n)*?)\\2","gm");
+const attrsRegx = new RegExp('([^\\s=]+)\\s*(=\\s*([\'"])([\\s\\S]*?)\\3)?', 'gm');
+
+function buildAttributesMap(attrStr, jPath, tagName) {
+  if (this.options.ignoreAttributes !== true && typeof attrStr === 'string') {
+    // attrStr = attrStr.replace(/\r?\n/g, ' ');
+    //attrStr = attrStr || attrStr.trim();
+
+    const matches = getAllMatches(attrStr, attrsRegx);
+    const len = matches.length; //don't make it inline
+    const attrs = {};
+
+    // First pass: parse all attributes and update matcher with raw values
+    // This ensures the matcher has all attribute values when processors run
+    const rawAttrsForMatcher = {};
+    for (let i = 0; i < len; i++) {
+      const attrName = this.resolveNameSpace(matches[i][1]);
+      const oldVal = matches[i][4];
+
+      if (attrName.length && oldVal !== undefined) {
+        let parsedVal = oldVal;
+        if (this.options.trimValues) {
+          parsedVal = parsedVal.trim();
+        }
+        parsedVal = this.replaceEntitiesValue(parsedVal, tagName, jPath);
+        rawAttrsForMatcher[attrName] = parsedVal;
+      }
+    }
+
+    // Update matcher with raw attribute values BEFORE running processors
+    if (Object.keys(rawAttrsForMatcher).length > 0 && typeof jPath === 'object' && jPath.updateCurrent) {
+      jPath.updateCurrent(rawAttrsForMatcher);
+    }
+
+    // Second pass: now process attributes with matcher having full attribute context
+    for (let i = 0; i < len; i++) {
+      const attrName = this.resolveNameSpace(matches[i][1]);
+
+      // Convert jPath to string if needed for ignoreAttributesFn
+      const jPathStr = this.options.jPath ? jPath.toString() : jPath;
+      if (this.ignoreAttributesFn(attrName, jPathStr)) {
+        continue
+      }
+
+      let oldVal = matches[i][4];
+      let aName = this.options.attributeNamePrefix + attrName;
+
+      if (attrName.length) {
+        if (this.options.transformAttributeName) {
+          aName = this.options.transformAttributeName(aName);
+        }
+        //if (aName === "__proto__") aName = "#__proto__";
+        aName = sanitizeName(aName, this.options);
+
+        if (oldVal !== undefined) {
+          if (this.options.trimValues) {
+            oldVal = oldVal.trim();
+          }
+          oldVal = this.replaceEntitiesValue(oldVal, tagName, jPath);
+
+          // Pass jPath string or matcher based on options.jPath setting
+          const jPathOrMatcher = this.options.jPath ? jPath.toString() : jPath;
+          const newVal = this.options.attributeValueProcessor(attrName, oldVal, jPathOrMatcher);
+          if (newVal === null || newVal === undefined) {
+            //don't parse
+            attrs[aName] = oldVal;
+          } else if (typeof newVal !== typeof oldVal || newVal !== oldVal) {
+            //overwrite
+            attrs[aName] = newVal;
+          } else {
+            //parse
+            attrs[aName] = parseValue(
+              oldVal,
+              this.options.parseAttributeValue,
+              this.options.numberParseOptions
+            );
+          }
+        } else if (this.options.allowBooleanAttributes) {
+          attrs[aName] = true;
+        }
+      }
+    }
+
+    if (!Object.keys(attrs).length) {
+      return;
+    }
+    if (this.options.attributesGroupName) {
+      const attrCollection = {};
+      attrCollection[this.options.attributesGroupName] = attrs;
+      return attrCollection;
+    }
+    return attrs
+  }
+}
+
+const parseXml = function (xmlData) {
+  xmlData = xmlData.replace(/\r\n?/g, "\n"); //TODO: remove this line
+  const xmlObj = new xmlNode('!xml');
+  let currentNode = xmlObj;
+  let textData = "";
+
+  // Reset matcher for new document
+  this.matcher.reset();
+
+  // Reset entity expansion counters for this document
+  this.entityExpansionCount = 0;
+  this.currentExpandedLength = 0;
+
+  const docTypeReader = new DocTypeReader(this.options.processEntities);
+  for (let i = 0; i < xmlData.length; i++) {//for each char in XML data
+    const ch = xmlData[i];
+    if (ch === '<') {
+      // const nextIndex = i+1;
+      // const _2ndChar = xmlData[nextIndex];
+      if (xmlData[i + 1] === '/') {//Closing Tag
+        const closeIndex = findClosingIndex(xmlData, ">", i, "Closing Tag is not closed.")
+        let tagName = xmlData.substring(i + 2, closeIndex).trim();
+
+        if (this.options.removeNSPrefix) {
+          const colonIndex = tagName.indexOf(":");
+          if (colonIndex !== -1) {
+            tagName = tagName.substr(colonIndex + 1);
+          }
+        }
+
+        tagName = transformTagName(this.options.transformTagName, tagName, "", this.options).tagName;
+
+        if (currentNode) {
+          textData = this.saveTextToParentTag(textData, currentNode, this.matcher);
+        }
+
+        //check if last tag of nested tag was unpaired tag
+        const lastTagName = this.matcher.getCurrentTag();
+        if (tagName && this.options.unpairedTags.indexOf(tagName) !== -1) {
+          throw new Error(`Unpaired tag can not be used as closing tag: </${tagName}>`);
+        }
+        if (lastTagName && this.options.unpairedTags.indexOf(lastTagName) !== -1) {
+          // Pop the unpaired tag
+          this.matcher.pop();
+          this.tagsNodeStack.pop();
+        }
+        // Pop the closing tag
+        this.matcher.pop();
+        this.isCurrentNodeStopNode = false; // Reset flag when closing tag
+
+        currentNode = this.tagsNodeStack.pop();//avoid recursion, set the parent tag scope
+        textData = "";
+        i = closeIndex;
+      } else if (xmlData[i + 1] === '?') {
+
+        let tagData = readTagExp(xmlData, i, false, "?>");
+        if (!tagData) throw new Error("Pi Tag is not closed.");
+
+        textData = this.saveTextToParentTag(textData, currentNode, this.matcher);
+        if ((this.options.ignoreDeclaration && tagData.tagName === "?xml") || this.options.ignorePiTags) {
+          //do nothing
+        } else {
+
+          const childNode = new xmlNode(tagData.tagName);
+          childNode.add(this.options.textNodeName, "");
+
+          if (tagData.tagName !== tagData.tagExp && tagData.attrExpPresent) {
+            childNode[":@"] = this.buildAttributesMap(tagData.tagExp, this.matcher, tagData.tagName);
+          }
+          this.addChild(currentNode, childNode, this.matcher, i);
+        }
+
+
+        i = tagData.closeIndex + 1;
+      } else if (xmlData.substr(i + 1, 3) === '!--') {
+        const endIndex = findClosingIndex(xmlData, "-->", i + 4, "Comment is not closed.")
+        if (this.options.commentPropName) {
+          const comment = xmlData.substring(i + 4, endIndex - 2);
+
+          textData = this.saveTextToParentTag(textData, currentNode, this.matcher);
+
+          currentNode.add(this.options.commentPropName, [{ [this.options.textNodeName]: comment }]);
+        }
+        i = endIndex;
+      } else if (xmlData.substr(i + 1, 2) === '!D') {
+        const result = docTypeReader.readDocType(xmlData, i);
+        this.docTypeEntities = result.entities;
+        i = result.i;
+      } else if (xmlData.substr(i + 1, 2) === '![') {
+        const closeIndex = findClosingIndex(xmlData, "]]>", i, "CDATA is not closed.") - 2;
+        const tagExp = xmlData.substring(i + 9, closeIndex);
+
+        textData = this.saveTextToParentTag(textData, currentNode, this.matcher);
+
+        let val = this.parseTextData(tagExp, currentNode.tagname, this.matcher, true, false, true, true);
+        if (val == undefined) val = "";
+
+        //cdata should be set even if it is 0 length string
+        if (this.options.cdataPropName) {
+          currentNode.add(this.options.cdataPropName, [{ [this.options.textNodeName]: tagExp }]);
+        } else {
+          currentNode.add(this.options.textNodeName, val);
+        }
+
+        i = closeIndex + 2;
+      } else {//Opening tag
+        let result = readTagExp(xmlData, i, this.options.removeNSPrefix);
+
+        // Safety check: readTagExp can return undefined
+        if (!result) {
+          // Log context for debugging
+          const context = xmlData.substring(Math.max(0, i - 50), Math.min(xmlData.length, i + 50));
+          throw new Error(`readTagExp returned undefined at position ${i}. Context: "${context}"`);
+        }
+
+        let tagName = result.tagName;
+        const rawTagName = result.rawTagName;
+        let tagExp = result.tagExp;
+        let attrExpPresent = result.attrExpPresent;
+        let closeIndex = result.closeIndex;
+
+        ({ tagName, tagExp } = transformTagName(this.options.transformTagName, tagName, tagExp, this.options));
+
+        if (this.options.strictReservedNames &&
+          (tagName === this.options.commentPropName
+            || tagName === this.options.cdataPropName
+          )) {
+          throw new Error(`Invalid tag name: ${tagName}`);
+        }
+
+        //save text as child node
+        if (currentNode && textData) {
+          if (currentNode.tagname !== '!xml') {
+            //when nested tag is found
+            textData = this.saveTextToParentTag(textData, currentNode, this.matcher, false);
+          }
+        }
+
+        //check if last tag was unpaired tag
+        const lastTag = currentNode;
+        if (lastTag && this.options.unpairedTags.indexOf(lastTag.tagname) !== -1) {
+          currentNode = this.tagsNodeStack.pop();
+          this.matcher.pop();
+        }
+
+        // Clean up self-closing syntax BEFORE processing attributes
+        // This is where tagExp gets the trailing / removed
+        let isSelfClosing = false;
+        if (tagExp.length > 0 && tagExp.lastIndexOf("/") === tagExp.length - 1) {
+          isSelfClosing = true;
+          if (tagName[tagName.length - 1] === "/") {
+            tagName = tagName.substr(0, tagName.length - 1);
+            tagExp = tagName;
+          } else {
+            tagExp = tagExp.substr(0, tagExp.length - 1);
+          }
+
+          // Re-check attrExpPresent after cleaning
+          attrExpPresent = (tagName !== tagExp);
+        }
+
+        // Now process attributes with CLEAN tagExp (no trailing /)
+        let prefixedAttrs = null;
+        let rawAttrs = {};
+        let namespace = undefined;
+
+        // Extract namespace from rawTagName
+        namespace = extractNamespace(rawTagName);
+
+        // Push tag to matcher FIRST (with empty attrs for now) so callbacks see correct path
+        if (tagName !== xmlObj.tagname) {
+          this.matcher.push(tagName, {}, namespace);
+        }
+
+        // Now build attributes - callbacks will see correct matcher state
+        if (tagName !== tagExp && attrExpPresent) {
+          // Build attributes (returns prefixed attributes for the tree)
+          // Note: buildAttributesMap now internally updates the matcher with raw attributes
+          prefixedAttrs = this.buildAttributesMap(tagExp, this.matcher, tagName);
+
+          if (prefixedAttrs) {
+            // Extract raw attributes (without prefix) for our use
+            rawAttrs = extractRawAttributes(prefixedAttrs, this.options);
+          }
+        }
+
+        // Now check if this is a stop node (after attributes are set)
+        if (tagName !== xmlObj.tagname) {
+          this.isCurrentNodeStopNode = this.isItStopNode(this.stopNodeExpressions, this.matcher);
+        }
+
+        const startIndex = i;
+        if (this.isCurrentNodeStopNode) {
+          let tagContent = "";
+
+          // For self-closing tags, content is empty
+          if (isSelfClosing) {
+            i = result.closeIndex;
+          }
+          //unpaired tag
+          else if (this.options.unpairedTags.indexOf(tagName) !== -1) {
+            i = result.closeIndex;
+          }
+          //normal tag
+          else {
+            //read until closing tag is found
+            const result = this.readStopNodeData(xmlData, rawTagName, closeIndex + 1);
+            if (!result) throw new Error(`Unexpected end of ${rawTagName}`);
+            i = result.i;
+            tagContent = result.tagContent;
+          }
+
+          const childNode = new xmlNode(tagName);
+
+          if (prefixedAttrs) {
+            childNode[":@"] = prefixedAttrs;
+          }
+
+          // For stop nodes, store raw content as-is without any processing
+          childNode.add(this.options.textNodeName, tagContent);
+
+          this.matcher.pop(); // Pop the stop node tag
+          this.isCurrentNodeStopNode = false; // Reset flag
+
+          this.addChild(currentNode, childNode, this.matcher, startIndex);
+        } else {
+          //selfClosing tag
+          if (isSelfClosing) {
+            ({ tagName, tagExp } = transformTagName(this.options.transformTagName, tagName, tagExp, this.options));
+
+            const childNode = new xmlNode(tagName);
+            if (prefixedAttrs) {
+              childNode[":@"] = prefixedAttrs;
+            }
+            this.addChild(currentNode, childNode, this.matcher, startIndex);
+            this.matcher.pop(); // Pop self-closing tag
+            this.isCurrentNodeStopNode = false; // Reset flag
+          }
+          else if (this.options.unpairedTags.indexOf(tagName) !== -1) {//unpaired tag
+            const childNode = new xmlNode(tagName);
+            if (prefixedAttrs) {
+              childNode[":@"] = prefixedAttrs;
+            }
+            this.addChild(currentNode, childNode, this.matcher, startIndex);
+            this.matcher.pop(); // Pop unpaired tag
+            this.isCurrentNodeStopNode = false; // Reset flag
+            i = result.closeIndex;
+            // Continue to next iteration without changing currentNode
+            continue;
+          }
+          //opening tag
+          else {
+            const childNode = new xmlNode(tagName);
+            if (this.tagsNodeStack.length > this.options.maxNestedTags) {
+              throw new Error("Maximum nested tags exceeded");
+            }
+            this.tagsNodeStack.push(currentNode);
+
+            if (prefixedAttrs) {
+              childNode[":@"] = prefixedAttrs;
+            }
+            this.addChild(currentNode, childNode, this.matcher, startIndex);
+            currentNode = childNode;
+          }
+          textData = "";
+          i = closeIndex;
+        }
+      }
+    } else {
+      textData += xmlData[i];
+    }
+  }
+  return xmlObj.child;
+}
+
+function addChild(currentNode, childNode, matcher, startIndex) {
+  // unset startIndex if not requested
+  if (!this.options.captureMetaData) startIndex = undefined;
+
+  // Pass jPath string or matcher based on options.jPath setting
+  const jPathOrMatcher = this.options.jPath ? matcher.toString() : matcher;
+  const result = this.options.updateTag(childNode.tagname, jPathOrMatcher, childNode[":@"])
+  if (result === false) {
+    //do nothing
+  } else if (typeof result === "string") {
+    childNode.tagname = result
+    currentNode.addChild(childNode, startIndex);
+  } else {
+    currentNode.addChild(childNode, startIndex);
+  }
+}
+
+/**
+ * @param {object} val - Entity object with regex and val properties
+ * @param {string} tagName - Tag name
+ * @param {string|Matcher} jPath - jPath string or Matcher instance based on options.jPath
+ */
+function replaceEntitiesValue(val, tagName, jPath) {
+  const entityConfig = this.options.processEntities;
+
+  if (!entityConfig || !entityConfig.enabled) {
+    return val;
+  }
+
+  // Check if tag is allowed to contain entities
+  if (entityConfig.allowedTags) {
+    const jPathOrMatcher = this.options.jPath ? jPath.toString() : jPath;
+    const allowed = Array.isArray(entityConfig.allowedTags)
+      ? entityConfig.allowedTags.includes(tagName)
+      : entityConfig.allowedTags(tagName, jPathOrMatcher);
+
+    if (!allowed) {
+      return val;
+    }
+  }
+
+  // Apply custom tag filter if provided
+  if (entityConfig.tagFilter) {
+    const jPathOrMatcher = this.options.jPath ? jPath.toString() : jPath;
+    if (!entityConfig.tagFilter(tagName, jPathOrMatcher)) {
+      return val; // Skip based on custom filter
+    }
+  }
+
+  // Replace DOCTYPE entities
+  for (const entityName of Object.keys(this.docTypeEntities)) {
+    const entity = this.docTypeEntities[entityName];
+    const matches = val.match(entity.regx);
+
+    if (matches) {
+      // Track expansions
+      this.entityExpansionCount += matches.length;
+
+      // Check expansion limit
+      if (entityConfig.maxTotalExpansions &&
+        this.entityExpansionCount > entityConfig.maxTotalExpansions) {
+        throw new Error(
+          `Entity expansion limit exceeded: ${this.entityExpansionCount} > ${entityConfig.maxTotalExpansions}`
+        );
+      }
+
+      // Store length before replacement
+      const lengthBefore = val.length;
+      val = val.replace(entity.regx, entity.val);
+
+      // Check expanded length immediately after replacement
+      if (entityConfig.maxExpandedLength) {
+        this.currentExpandedLength += (val.length - lengthBefore);
+
+        if (this.currentExpandedLength > entityConfig.maxExpandedLength) {
+          throw new Error(
+            `Total expanded content size exceeded: ${this.currentExpandedLength} > ${entityConfig.maxExpandedLength}`
+          );
+        }
+      }
+    }
+  }
+  // Replace standard entities
+  for (const entityName of Object.keys(this.lastEntities)) {
+    const entity = this.lastEntities[entityName];
+    const matches = val.match(entity.regex);
+    if (matches) {
+      this.entityExpansionCount += matches.length;
+      if (entityConfig.maxTotalExpansions &&
+        this.entityExpansionCount > entityConfig.maxTotalExpansions) {
+        throw new Error(
+          `Entity expansion limit exceeded: ${this.entityExpansionCount} > ${entityConfig.maxTotalExpansions}`
+        );
+      }
+    }
+    val = val.replace(entity.regex, entity.val);
+  }
+  if (val.indexOf('&') === -1) return val;
+
+  // Replace HTML entities if enabled
+  if (this.options.htmlEntities) {
+    for (const entityName of Object.keys(this.htmlEntities)) {
+      const entity = this.htmlEntities[entityName];
+      const matches = val.match(entity.regex);
+      if (matches) {
+        //console.log(matches);
+        this.entityExpansionCount += matches.length;
+        if (entityConfig.maxTotalExpansions &&
+          this.entityExpansionCount > entityConfig.maxTotalExpansions) {
+          throw new Error(
+            `Entity expansion limit exceeded: ${this.entityExpansionCount} > ${entityConfig.maxTotalExpansions}`
+          );
+        }
+      }
+      val = val.replace(entity.regex, entity.val);
+    }
+  }
+
+  // Replace ampersand entity last
+  val = val.replace(this.ampEntity.regex, this.ampEntity.val);
+
+  return val;
+}
+
+
+function saveTextToParentTag(textData, parentNode, matcher, isLeafNode) {
+  if (textData) { //store previously collected data as textNode
+    if (isLeafNode === undefined) isLeafNode = parentNode.child.length === 0
+
+    textData = this.parseTextData(textData,
+      parentNode.tagname,
+      matcher,
+      false,
+      parentNode[":@"] ? Object.keys(parentNode[":@"]).length !== 0 : false,
+      isLeafNode);
+
+    if (textData !== undefined && textData !== "")
+      parentNode.add(this.options.textNodeName, textData);
+    textData = "";
+  }
+  return textData;
+}
+
+//TODO: use jPath to simplify the logic
+/**
+ * @param {Array<Expression>} stopNodeExpressions - Array of compiled Expression objects
+ * @param {Matcher} matcher - Current path matcher
+ */
+function isItStopNode(stopNodeExpressions, matcher) {
+  if (!stopNodeExpressions || stopNodeExpressions.length === 0) return false;
+
+  for (let i = 0; i < stopNodeExpressions.length; i++) {
+    if (matcher.matches(stopNodeExpressions[i])) {
+      return true;
+    }
+  }
+  return false;
+}
+
+/**
+ * Returns the tag Expression and where it is ending handling single-double quotes situation
+ * @param {string} xmlData 
+ * @param {number} i starting index
+ * @returns 
+ */
+function tagExpWithClosingIndex(xmlData, i, closingChar = ">") {
+  let attrBoundary;
+  let tagExp = "";
+  for (let index = i; index < xmlData.length; index++) {
+    let ch = xmlData[index];
+    if (attrBoundary) {
+      if (ch === attrBoundary) attrBoundary = "";//reset
+    } else if (ch === '"' || ch === "'") {
+      attrBoundary = ch;
+    } else if (ch === closingChar[0]) {
+      if (closingChar[1]) {
+        if (xmlData[index + 1] === closingChar[1]) {
+          return {
+            data: tagExp,
+            index: index
+          }
+        }
+      } else {
+        return {
+          data: tagExp,
+          index: index
+        }
+      }
+    } else if (ch === '\t') {
+      ch = " "
+    }
+    tagExp += ch;
+  }
+}
+
+function findClosingIndex(xmlData, str, i, errMsg) {
+  const closingIndex = xmlData.indexOf(str, i);
+  if (closingIndex === -1) {
+    throw new Error(errMsg)
+  } else {
+    return closingIndex + str.length - 1;
+  }
+}
+
+function readTagExp(xmlData, i, removeNSPrefix, closingChar = ">") {
+  const result = tagExpWithClosingIndex(xmlData, i + 1, closingChar);
+  if (!result) return;
+  let tagExp = result.data;
+  const closeIndex = result.index;
+  const separatorIndex = tagExp.search(/\s/);
+  let tagName = tagExp;
+  let attrExpPresent = true;
+  if (separatorIndex !== -1) {//separate tag name and attributes expression
+    tagName = tagExp.substring(0, separatorIndex);
+    tagExp = tagExp.substring(separatorIndex + 1).trimStart();
+  }
+
+  const rawTagName = tagName;
+  if (removeNSPrefix) {
+    const colonIndex = tagName.indexOf(":");
+    if (colonIndex !== -1) {
+      tagName = tagName.substr(colonIndex + 1);
+      attrExpPresent = tagName !== result.data.substr(colonIndex + 1);
+    }
+  }
+
+  return {
+    tagName: tagName,
+    tagExp: tagExp,
+    closeIndex: closeIndex,
+    attrExpPresent: attrExpPresent,
+    rawTagName: rawTagName,
+  }
+}
+/**
+ * find paired tag for a stop node
+ * @param {string} xmlData 
+ * @param {string} tagName 
+ * @param {number} i 
+ */
+function readStopNodeData(xmlData, tagName, i) {
+  const startIndex = i;
+  // Starting at 1 since we already have an open tag
+  let openTagCount = 1;
+
+  for (; i < xmlData.length; i++) {
+    if (xmlData[i] === "<") {
+      if (xmlData[i + 1] === "/") {//close tag
+        const closeIndex = findClosingIndex(xmlData, ">", i, `${tagName} is not closed`);
+        let closeTagName = xmlData.substring(i + 2, closeIndex).trim();
+        if (closeTagName === tagName) {
+          openTagCount--;
+          if (openTagCount === 0) {
+            return {
+              tagContent: xmlData.substring(startIndex, i),
+              i: closeIndex
+            }
+          }
+        }
+        i = closeIndex;
+      } else if (xmlData[i + 1] === '?') {
+        const closeIndex = findClosingIndex(xmlData, "?>", i + 1, "StopNode is not closed.")
+        i = closeIndex;
+      } else if (xmlData.substr(i + 1, 3) === '!--') {
+        const closeIndex = findClosingIndex(xmlData, "-->", i + 3, "StopNode is not closed.")
+        i = closeIndex;
+      } else if (xmlData.substr(i + 1, 2) === '![') {
+        const closeIndex = findClosingIndex(xmlData, "]]>", i, "StopNode is not closed.") - 2;
+        i = closeIndex;
+      } else {
+        const tagData = readTagExp(xmlData, i, '>')
+
+        if (tagData) {
+          const openTagName = tagData && tagData.tagName;
+          if (openTagName === tagName && tagData.tagExp[tagData.tagExp.length - 1] !== "/") {
+            openTagCount++;
+          }
+          i = tagData.closeIndex;
+        }
+      }
+    }
+  }//end for loop
+}
+
+function parseValue(val, shouldParse, options) {
+  if (shouldParse && typeof val === 'string') {
+    //console.log(options)
+    const newval = val.trim();
+    if (newval === 'true') return true;
+    else if (newval === 'false') return false;
+    else return toNumber(val, options);
+  } else {
+    if (isExist(val)) {
+      return val;
+    } else {
+      return '';
+    }
+  }
+}
+
+function fromCodePoint(str, base, prefix) {
+  const codePoint = Number.parseInt(str, base);
+
+  if (codePoint >= 0 && codePoint <= 0x10FFFF) {
+    return String.fromCodePoint(codePoint);
+  } else {
+    return prefix + str + ";";
+  }
+}
+
+function transformTagName(fn, tagName, tagExp, options) {
+  if (fn) {
+    const newTagName = fn(tagName);
+    if (tagExp === tagName) {
+      tagExp = newTagName
+    }
+    tagName = newTagName;
+  }
+  tagName = sanitizeName(tagName, options);
+  return { tagName, tagExp };
+}
+
+
+
+function sanitizeName(name, options) {
+  if (criticalProperties.includes(name)) {
+    throw new Error(`[SECURITY] Invalid name: "${name}" is a reserved JavaScript keyword that could cause prototype pollution`);
+  } else if (DANGEROUS_PROPERTY_NAMES.includes(name)) {
+    return options.onDangerousProperty(name);
+  }
+  return name;
+}